Darlene King, MD

Clinicians Encounter Patient Use of AI for Mental Health, With Darlene King, MD

As patients increasingly turn to artificial intelligence (AI) tools for 

guidance, clinicians are encountering new dynamics in clinical encounters.

Darlene King, MD, assistant professor in the department of psychiatry at UT Southwestern Medical Center, recalled a time when a patient causally pulled out ChatGPT during a psychiatric visit.

“They just asked ChatGPT in front of me what it thought about the medication I was offering,” King told 

Other times, King had patients send her chart messages that would say that they asked ChatGPT about their medication plan and this was the response.

“They'll give me a whole printout of their conversation with ChatGPT, and I’ll read through it and talk about it with them, usually,” she said.

King noted that AI-generated responses occasionally align with clinical plans, which can reinforce treatment decisions and support patient understanding. However, she also described scenarios in which AI-generated content conflicts with established diagnoses.

“I’ve had them tell me ChatGPT has known me for 2 years, and it doesn’t think I have schizophrenia,” King said. “It thinks I have X, Y or Z instead.”

In this case, she said there needs to be careful discussion with the patient to review symptoms and prior clinical assessments.

King emphasized that clinicians should proactively ask about AI use during routine history taking. Understanding how frequently patients rely on these tools and for what purposes can help identify potential concerns, including excessive use or reliance in place of social support. She also highlighted privacy considerations, noting that patients may not recognize that AI platforms are not confidential.

Although she has not observed AI directly worsening symptoms, King said it can reinforce preexisting beliefs that diverge from clinical recommendations. “It’s kind of like you go into ChatGPT with…something you want to hear… and it can amplify that,” she said.

King views AI’s role in psychiatry as uncertain, with potential to both augment care and disrupt traditional models, particularly as direct-to-consumer mental health tools expand. She underscored the need for stronger evidence and regulatory frameworks to guide safe implementation.

“Ideally it would help augment care,” King said, “[but] we really need to study and get some good evidence to say whether it can be safe in new ways.”

Watch the full dynamic feature on using AI as therapist, featuring King, 

Videos

King discusses real-world patient use of AI tools for mental health advice and implications for clinical conversations.

Your Patient's New Therapist Is an AI. Now What?

Watch the full expert dynamic feature above.

In your clinic, you may hear patients say, “Chat told me this” or “Chat told me that.” Numerous individuals use ChatGPT, colloquially referred to as “Chat,” to get quick, convenient 

advice. This can be dangerous, depending on what the chatbot says, potentially providing patients with advice you would never, in a million years, give.

“More often, it's actually deleterious where information is cherry-picked, and the context is not always understood,” said Richard Miller, MD, a staff psychiatrist at Elwyn Adult Behavioral Health in Cranston, RI. “Patients are kind of going on an extended trip…to get the information that they're looking for…. maybe [a] particular answer. It's not always accurate.”

Despite the danger, AI can also be helpful in the psychiatry field. AI can provide a new perspective and can augment, not replace, treatment.

 Editorial Team put together a feature examining the benefits, dangers, and ethics surrounding AI use in psychiatry. Millions of Americans struggle to access mental health care, with waitlists sometimes lasting 3 to 6 months.

These individuals, desperate for mental health advice, turn to AI. As Darlene King, MD, from the University of Texas Southwest Medical Center, said, using AI as a therapist is like getting mental health advice from a friend, who can very well give bad advice.

Dominic Sisti, PhD, from Penn Medicine, discussed the ethical concerns of AI in psychiatry. Vulnerable populations may be susceptible to abusing AI and can even go as far as asking ChatGPT ways to self-harm. “How would these AI platforms [respond] to a patient who is indicating they have suicidal ideation or behavior?” he asked. “How might these platforms recognize elevated risk for self-harm or harm to others?

Sisti stressed the need for guardrails on age restrictions, redirection protocols, and HIPAA compliance standards. He also said there is a need for platforms to distinguish between an investigator asking about suicide methods versus someone in crisis.

Other than mental health advice, AI can assist in screening. In 2024, a study showed the promise of Sonde Health’s mental fitness vocal biomarker tool to 

identify depression from the sound of a voice

 Another study showed the potential of MoodCapture, a smartphone application using AI to 

detect the onset of depression based on facial cues 

More recently, a study showed AI can predict treatment response in major depressive disorder (MDD). In the 

phase 2b OLIVE trial assessing BH-200 (nelivaptan)

, a selective vasopressin V1b receptor antagonist, a proprietary genetic tool stratified patients by vasopressin-related biomarkers.

 Patients with lower peripheral but higher central vasopressin activity had a stronger antidepressant response, supporting biomarker-guided patient selection.

AI can also help with psychiatrists’ day-to-day responsibilities. For instance, Suki AI, Heidi, and DeepScribe, among others, are designed for clinician documentation.

As AI becomes more commonplace, experts recommend asking about its use non-judgmentally during routine history-taking. Although potentially dangerous, AI can be beneficial if used with a “human in the loop,” according to Sisti.

 Staff psychiatrist at Elwyn Adult Behavioral Health in Cranston, RI.

Associate professor of medical ethics & health policy, director of the scattergood program for applied ethics in behavioral health care, associate professor of psychiatry (secondary), associate professor of philosophy (secondary) at the University of Pennsylvania.

Assistant professor in the department of psychiatry at University of Texas Southwest Medical Center who specializes in general adult psychiatry, including women’s health, addiction, and trauma.

A psychiatrist and chief medical officer at HMNC Brain Health.

Relevant disclosures include Teva Pharmaceuticals USA, Inc., Alkermes Janssen Pharmaceuticals, Inc, and Otsuka America Pharmaceutical for MIller. King, Sisti and Eriksson have no reported disclosures.

Andrew. How Long Is the Wait to See a Psychiatrist? What to Expect and How to Shorten It. Brain Health USA. Published September 30, 2025. 

https://brainhealthusa.com/how-long-is-the-wait-to-see-a-psychiatrist/

Derman C. A Voice Detecting Depression? Lindsey Venesky, PhD, Discusses New Data. HCPLive. March 14, 2024. Accessed on March 30, 2026. 

https://www.hcplive.com/view/a-voice-detecting-depression-lindsey-venesky-phd-discusses-new-data

Derman, C. Smartphone App with AI Detects Depression Onset from Facial Expressions. HCPLive. February 28, 2024. Accessed March 30, 2026. 

https://www.hcplive.com/view/smartphone-app-ai-detects-depression-onset-facial-expressions

Eriksson H. New Genetic Tool Shows Promise for Targeted MDD Treatment, With Hans Eriksson, PhD, MD. HCPLive. Published December 5, 2026. Accessed March 30, 2026. 

https://www.hcplive.com/view/new-genetic-tool-shows-promise-targeted-mdd-treatment-hans-eriksson-md-phd

Palm E, Manikantan A, Mahal H, Belwadi SS, Pepin ME. Assessing the quality of AI-generated clinical notes: validated evaluation of a large language model ambient scribe. 

. 2025;8:1691499. Published 2025 Oct 22. 

Mess SA, Mackey AJ, Yarowsky DE. Artificial Intelligence Scribe and Large Language Model Technology in Healthcare Documentation: Advantages, Limitations, and Recommendations. 

. 2025;13(1):e6450. Published 2025 Jan 16. 

DeepScribe. DeepScribe Solidifies Ambient AI Leadership in Oncology with New Study and Accelerated Growth. Prnewswire.com. Published September 16, 2025. Accessed March 30, 2026. 

https://www.prnewswire.com/news-releases/deepscribe-solidifies-ambient-ai-leadership-in-oncology-with-new-study-and-accelerated-growth-302557045.html

As patients increasingly turn to ChatGPT for mental health support, psychiatrists and ethicists weigh the clinical risks, ethical concerns, and what clinicians should do next.

Darlene King, MD

Articles